A hybrid deep learning-based intrusion detection system for IoT networks.

Area of Science:

• Cybersecurity
• Network Security
• Deep Learning Applications

Background:

• Internet of Things (IoT) networks face significant security challenges, with existing intrusion detection systems (IDS) often limited to single-layer analysis.
• The need for comprehensive, multi-layered intrusion detection in IoT environments is critical due to the evolving threat landscape.

Purpose of the Study:

• To propose and evaluate an intelligent IDS for IoT networks capable of detecting intrusions across physical, network, and application layers.
• To leverage deep learning, specifically Recurrent Neural Network-Gated Recurrent Units (RNN-GRU), for enhanced multi-layer attack classification.

Main Methods:

• Development of a novel deep learning model (RNN-GRU) for classifying attacks across the three layers of IoT architecture.
• Training and testing the model using the specialized ToN-IoT dataset, which includes novel attack vectors.
• Performance evaluation using metrics like accuracy, precision, recall, and F1-score, with Adam optimization.

Main Results:

• The proposed RNN-GRU model achieved high accuracy: 99% for network flow datasets and 98% for application layer datasets.
• Demonstrated superior performance compared to various advanced deep learning and traditional machine learning techniques.
• The Adam optimizer proved optimal for model evaluation.

Conclusions:

• The developed intelligent IDS effectively addresses the limitations of single-layer detection systems in IoT.
• The RNN-GRU model offers a robust and superior solution for multi-layered intrusion detection in IoT networks.
• This research contributes a significant advancement in securing the Internet of Things.